Fault Tolerance through Redundancy

Introduction

The term "redundancy," as used in this guide, is the ability to configure one or more backup components (or cards) to take over for a component that fails. Redundancy provides system-wide fault tolerance, helping to ensure that the CSP continues to process calls despite a hardware or software fault.

The CSP has a dual bus structure and dual host connections. It can be configured with dual power supplies, and it supports redundant Matrix Controller cards. The CSP also supports N+1 configuration for both line cards and ISDN cards.

Redundant Matrix Controller Cards

In a CSP with redundant Matrix Controller cards, the cards themselves are physically identical. It is only their designation by the host, through a process known as "hardware arbitration," that determines which is "active" and which is "standby." The Matrix Controller card designated "active" has full access to all hardware resources, while the card designated "standby" is ready to assume all functions of the active Matrix Controller card under the following conditions:

• The host requests a switchover.

• Software is downloaded to the switch.

• The active Matrix Controller card is removed or physically reset.

• The active Matrix Controller card is reconfigured.

• A hardware or software failure occurs after the system starts up and stabilizes.

Whenever the "standby" card takes over for the "active" card, it becomes the new "active" card, and is recognized as such by the CSP and host.

When system software is downloaded to one of the Matrix Controller cards, the host assigns a timestamp. The timestamp determines which Matrix Controller has the most recently downloaded system software. The system software with the most recent timestamp becomes resident and runs on both Matrix Controller cards.

The two Matrix Controller cards communicate with each other over a reserved, High-Level Data Link Control (HDLC) link. Pulse Code Modulation (PCM) data travels back and forth on the midplane bus between the Matrix Controller cards and line cards.

Redundant Line Cards

You can provide a CSP with N+1 redundancy for its line cards if you configure one extra line card, together with its I/O card, for each card type. N is the number of line cards of the same type in the CSP, and 1 is the one extra card that serves as a backup in the event of a hardware or software failure. Each redundant line card backs up several cards, but it can only back up a single card type.

For example, if your CSP has three T-ONE cards, you can insert a fourth T-ONE card to serve as a standby card in case one of the active T-ONE cards fails. If your CSP also has multiple E-ONE line
cards, you need an E-ONE card in standby mode to achieve E1 redundancy. When a line card fails, the Matrix Controller notifies the host of the failure and the host sends a Line Card Switchover (0x0024) message to the Matrix Controller to redirect signaling from the failed line card to the standby line card.

Important! Only one standby line card can switch to active mode at a time.

The Redundant I/O card redirects traffic to the redundant bus which moves traffic to the Standby I/O card. Note that standard I/O cards do not have this capability. The figure below depicts configuration of redundant line cards.

Figure 1-2 Configuration for Line Card Redundancy

Redundant Common Channel Signaling Cards

You can also configure redundancy for Common Channel Signaling (CCS) cards, which include SS7 and ISDN cards. Please see the Developer’s Guide: Common Channel Signaling book for more information.

Network Interfaces

The CSP supports the following network interfaces:

• E1

• T1

• J1

• DS3

Common Channel Signaling

The CSP supports the following common channel signaling:

• SS7

• ISDN PRI and BRI

• DASS2/DPNSS